A survey on intelligent routing in vehicular ad hoc networks using deep learning and gannet optimization

Abstract

Vehicular Ad Hoc Networks (VANETs) have emerged as an important communication technology for intelligent transportation systems, enabling vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-everything (V2X) communication. However, high vehicle mobility, rapidly changing network topology, intermittent connectivity, congestion, limited link lifetime, and stringent latency requirements make reliable routing a challenging problem. Conventional routing protocols are often unable to adapt rapidly to these dynamic conditions. Consequently, machine learning, deep learning, metaheuristic optimization, edge computing, and federated learning have increasingly been investigated for intelligent vehicular networking. Among metaheuristic techniques, the Gannet Optimization Algorithm (GOA) has attracted attention because of its exploration and exploitation mechanisms inspired by the foraging behavior of gannets. Recent research has demonstrated the applicability of GOA to VANET routing, while deep learning techniques such as graph convolutional networks, long short-term memory networks, graph neural networks, and attention mechanisms have been used for mobility prediction, congestion estimation, and intelligent routing. This paper surveys conventional VANET routing, optimization-based routing, GOA-based optimization, deep-learning-assisted routing, and edge/federated intelligence for vehicular networks. The comparative analysis identifies the strengths and limitations of existing approaches and highlights the lack of a unified framework that jointly exploits mobility-aware graph learning and adaptive multi-objective GOA for real-time routing. Based on this gap, a future research direction integrating graph learning, adaptive Gannet optimization, and edge intelligence is proposed.

Keywords